Research of coal mine water burst forecasting method based on PCA-ELM

In view of problems of slow training speed, poor generalization performance and low test precision existed in coal mine water burst forecasting model with traditional algorithm, the paper proposed a coal mine water burst forecasting method based on PCA-ELM , and established coal mine water burst forecasting model with the method. The method takes mine water burst historical data as sample, uses PCA to obtain mine water burst controlling factors, and divides the sample data containing the main controlling factors into train set, validation set and test set; then, takes the train set as input of ELM to train the model; finally, uses validation set to validate the model. The experimental results show that input variables of the method are less than the traditional methods, time of modeling and computing is shorter, operating speed and prediction accuracy of the model are higher.